Journal of Accident Investigation

Figure 2, illustrating control inputs during the second wake
turbulence encounter, shows that the first officer responded
to the second wake turbulence encounter much differently
than he did to the first. Wheel inputs during the second wake
turbulence encounter were about twice as large as those made
during the first, and the first officer also made rudder pedal
inputs. The NTSB found that the full wheel and rudder inputs
made in response to the second wake turbulence encounter
were unnecessary and excessive.
Figures 1 and 2 show individual parameters: column, wheel,
pedals, and time. However, it is not easy to understand the
magnitude of the data within the timeframe. In particular,
the pilot’s reaction to the first and second wake turbulence
encounters is hard to visualize using only the data in the figures.
Given these difficulties, staff decided to animate the first officer’s
input to make it easier to grasp the data. The following is the
detailed process used in the animation reconstruction:
• Storyboarded the first officer’s control inputs.
•
•
•
•
•
•
Modeled the human legs, pedals, and control column
according to the scale and measurements of the
A300-600 cockpit.
Wrote XSI script to read the CVR text file, and MS
Excel spreadsheet containing the FDR data of the
flight control inputs.
Textured the legs to show only their outlines, and
textured the cockpit components to represent the
A300-600 cockpit.
Synchronized digital time in accordance with flight
control input.
Reviewed data accuracy against the pedal and column
motion to ensure accurate representation.
Composited the data-driven control input and digital
time with selected cockpit communication (figures 3
and 4).
As the list above shows, translating field data to an animation
is not a single-step process. In addition to being factually
correct, the animation must demonstrate a high degree of data
analysis, conveyed so that the data are easy for an audience to
understand. In addition, the animation scene must be simple so
that the main focus is on the probable cause and the audience
is not distracted by extraneous factors. For these reasons, only
the control column and pedals were included for the cockpit
environment. However, the scene seemed incomplete without
a human figure, even though an entire human figure seemed to
distract attention from the control inputs. The solution was to
include only the lower torso of the figure, which worked well to
demonstrate the human movements without detracting from the
DEVELOPING ANIMATIONS TO SUPPORT COMPLEX AVIATION ACCIDENT INVESTIGATIONS
control inputs. Photographs, engineering drawings, and survey
data were used to accurately model the cockpit environment
and placement of the lower torso according to the actual
A300-600 cockpit. The green regions under the rudder pedals
were used to depict the range of available pedal travel prior to
reaching the pedal travel limits. (These limits correlate to the
grey lines in the pedal plots of figures 1 and 2.) The animation
also included digital time and selected cockpit communications
to round out the sequence of events.
Although extensive effort was required to animate the
sequence of flight control inputs, the effort was clearly
worthwhile since it demonstrates in real time the first officer’s
unnecessary and excessive rudder inputs during the wake
turbulence encounters and shows how his actions changed from
the first wake encounter to the second. Further, by showing
events in real time using the cockpit orientation, the animation
thoroughly represents the sequence of events and enhanced
investigators’ understanding of the control inputs. Finally, the
animation allows a nontechnical audience to watch the actions
associated with the control input data rather than viewing a
static diagram that must be explained by the presenter. Figures
1 and 2 show the data, but the animation (figures 3 and 4)
shows the human actions that resulted in that data. This 3-D
animation reconstruction effectively explains the first officer’s
actions and enables both the investigators and the audience to
visualize the complex control input data associated with the
accident.
Accident 2
On January 8, 2003, about 0847:28 eastern standard time, Air
Midwest flight 481, a Raytheon (Beechcraft) 1900D, crashed
shortly after taking off from Charlotte-Douglas International
Airport, Charlotte, North Carolina. Two flight crewmembers
and nineteen passengers aboard the airplane were killed, one
person on the ground received minor injuries, and impact forces
and a postcrash fire destroyed the airplane. NTSB determined
that the probable cause of the accident was the airplane’s
loss of pitch control during takeoff. The loss of pitch control
resulted from incorrect rigging of the elevator control system,
compounded by the airplane’s aft center of gravity, which was
substantially aft of the certified aft limit. []
Accident 2: Data Integration
The challenge in analyzing and presenting data from flight
481 was to explain how the airplane’s mis-rigged elevator
cable control system affected airplane motion, resulting in
loss of pitch control. Investigators found that they could not
effectively demonstrate the physical evidence associated with
the mis-rigged elevator control system or the airplane’s flight
NTSB JOURNAL OF ACCIDENT INVESTIGATION, SPRING 2006; VOLUME 2, ISSUE 1 2